We worked with professional racing teams to optimise driver performance. We studied laptimes, telemetry, team drivers, opponent drivers, driver cams, everything we could get. At the 12 Hours of Sebring, somewhere around hour four, we had it.
The Key Metric for professional auto racing.
You would think it would be fast laptimes. This is a reasonable assumption. The driver that gets around the racetrack the fastest is most likely to win, right? Not really. There's traffic. Other race cars. Drivers rarely get to stretch their legs on a track by themselves. Passing is where the real skill occurs in automotive racing. The drivers who can pass the other cars in the least amount of time are the drivers who are most likely to win.
We call it Latency to Pass. The time from when a driver first gets behind another car, to when it drives past that car. Lower latencies to pass mean they don't stay behind for long. It means they can overcome the traffic slowing them down.
Latency to pass was the Key Metric because it easily determined how to succeed. How to win races. It was an empirical, idemnotic measurement of human performance. And it resulted in more sponsorship funding for the teams that used this metric to select drivers.
We worked with trial attorneys at a small law firm. They wanted to use behavior science to help them win court cases.
We observed several cases from start to finish, from jury selection to court trials. We deconstructed the process with systems analyses. Identified controlling variables. Implemented metrics to guide successful action.
Typically, law firms attempt to select and exclude jurors by interviewing them to identify pre-existing biases or motivations. Hopefully, discovering this background information will help them select jurors that are likely to decide in their favor. Unfortunately, this doesn't really work. Jurors often lie or hide their true motivations. They may want to avoid jury duty, or participate and sway the outcome of the trial.
Instead, we took a first principles approach. What is really going on?
To win court cases, the jury must decide a verdict in favor of the attorney's client.
More specifically, the attorney's verbal behavior must affect the verbal behavior of the jurors.
Could we measure this effect?
Yes. During the jury selection phase, the attorney would ask for a potential juror's opinion on a controversial subject. We then measured the latency of answering, or how long it takes the juror to start speaking after the question is posed.
For example, the attorney might ask, "what is your opinion on theft?" The person might respond quickly, saying "It's bad, I'm against stealing." Say the latency of answering is 0.5 seconds.
The attorney then attempts to change their mind by providing a convincing argument. For example, they tell a story about a mother of a starving family stealing food for her children.
The attorney then repeats the original question. "How do you feel about stealing?" And we measure the latency again. Perhaps the person thinks for a moment, and then responds "Well...I guess stealing is bad...but maybe not in every situation." The latency is now 2 seconds, or a 400% increase. The effect size of the attorney's arugment is 400%.
We have now quantified the ability to change someone's mind.
In behavior science, short latencies mean the behavior is highly probable and resistant to change. Increasing the latency indicates a weakening of that response probability.
We started selecting jurors according to the magnitude of effect size. The larger the increase in latency, the more likely a juror could be convinced. In this example, the effect size is 400%. Across dozens of potential jurors, effect sizes ranged from -0.6 seconds to over two minutes. About -50% to over 2,400%. We rejected potential jurors with small effect sizes.
Did it work? Resoundingly, yes. The firm's trial win/loss ratio increased from 3/10 to 8/10.
Latency of answering questions was a Key Metric for winning court cases. It is empricial, idemnotic, and crystal clear. It guided successful action for the law firm.